An efficient load balancing multicast scheduling for solving congestion problem in social data center networks

Recently, social network sites have become more popular. The web site traffic on Facebook reaches 22.36% proportion of global data traffic in the world. There are various and complex kinds of data types such as texts, photographs, and videos which are transmitted in social network sites. In social networks, cloud services are generally accomplished by multicast-based group communications. The extensive data of social networks is generated within a relatively short period of time and is concentrated on partial servers. Subsequently, the rate of multicast congestion increases substantially, resulting in severe packet loss and transmission error. Therefore, we study the congestion problem of multicast-based group communications in the social data center network. Then, we propose an efficient load balancing multicast scheduling (LBMS) by observing users' behaviors on the social network to alleviate the congestion problems of the multicast traffic. Simulation results shows that LBMS can achieve load balance and significantly improve throughput and average delay.